US2023418596A1PendingUtilityA1
Automatically generating human language code descriptions
Est. expiryJun 23, 2042(~15.9 yrs left)· nominal 20-yr term from priority
G06N 3/08G06N 3/045G06F 8/73G06F 21/563G06F 21/577G06F 8/74G06F 11/3636G06F 11/302
47
PatentIndex Score
0
Cited by
0
References
0
Claims
Abstract
Devices, systems, and methods for executable binary file description are provided. A method can include generating or identifying an execution trace for the executable binary file. The method can include operating, based on the execution trace, a transformer model that generates a human language program description of the execution trace.
Claims
exact text as granted — not AI-modifiedWhat is claimed is:
1 . A method for executable binary file description, the method comprising:
generating or identifying an execution trace for the executable binary file; and operating, based on the execution trace, a transformer model that generates a human language program description of the execution trace.
2 . The method of claim 1 , wherein the execution trace includes a program counter sequence and the transformer model includes a first embedder that generates an embedding representation of the program counter sequence.
3 . The method of claim 2 , wherein the execution trace further includes an executed instruction sequence and the transformer model includes a second embedder that generates an embedding representation of the executed instruction sequence.
4 . The method of claim 3 , wherein the executed instruction sequence includes operation codes for machine code instructions executed in executing the executable binary file.
5 . The method of claim 3 , wherein the execution trace further includes a state of relevant variables.
6 . The method of claim 5 , wherein the state of relevant variables includes values of variables input into instructions of the execution instruction sequence.
7 . The method of claim 1 , further comprising reverse engineering the executable binary file based on the human language program description.
8 . The method of claim 1 , further comprising training the transformer model using a training dataset comprising (execution trace, program description) pairs.
9 . The method of claim 8 , wherein the program description of the (execution trace, program description) pairs are comments from program code used to generate the executable binary file.
10 . The method of claim 9 , wherein the comments are function level comments, program level comments, or a combination thereof.
11 . A non-transitory machine-readable medium including instructions that, when executed by a machine, cause the machine to perform operations for executable binary file description, the operations comprising:
generating or identifying an execution trace for the executable binary file; and operating, based on the execution trace, a transformer model that generates a human language program description of the execution trace.
12 . The non-transitory machine-readable medium of claim 11 , wherein the execution trace includes a program counter sequence and the transformer model includes a first embedder that generates an embedding representation of the program counter sequence.
13 . The non-transitory machine-readable medium of claim 12 , wherein the execution trace further includes an executed instruction sequence and the transformer model includes a second embedder that generates an embedding representation of the executed instruction sequence.
14 . The non-transitory machine-readable medium of claim 13 , wherein the executed instruction sequence includes operation codes for machine code instructions executed in executing the executable binary file.
15 . The non-transitory machine-readable medium of claim 13 , wherein the execution trace further includes a state of relevant variables.
16 . The non-transitory machine-readable medium of claim 15 , wherein the state of relevant variables includes values of variables input into instructions of the execution instruction sequence.
17 . A system comprising:
processing circuitry; memory including instructions that, when executed by the processing circuitry, cause the processing circuitry to perform operations for executable binary file description, the operations comprising:
generating or identifying an execution trace for the executable binary file; and
operating, based on the execution trace, a transformer model that generates a human language program description of the execution trace.
18 . The system of claim 17 , wherein the operations further comprise reverse engineering the executable binary file based on the human language program description.
19 . The system of claim 17 , wherein the operations further comprise training the transformer model using a training dataset comprising (execution trace, program description) pairs.
20 . The system of claim 18 , wherein the program description of the (execution trace, program description) pairs are comments from program code used to generate the executable binary file, wherein the comments are function level comments, program level comments, or a combination thereof.Join the waitlist — get patent alerts
Track US2023418596A1 — get alerts on status changes and closely related new filings.
We store only your email — no account needed. See our privacy policy.